Add Calculation of Policy Impacts on GDP and Jobs

[jrissman]

This is a substantial upgrade to the economic calculations that will need to be done carefully. I may post more info on the approach later, but some of the main points are:

• We want the calculations to be implemented in Vensim (e.g. not feed EPS outputs into a different computer model), so the outputs are supported by the web interface and can be done on-the-fly for custom policy packages
• We need to handle both positive and negative cash flow changes. That is, if an entity saves money, what else does the entity spend it on (within a year), and how much does the entity save? Or if the entity earns less revenue (or has to spend more on fuel, etc.), what does the entity spend less on in response?
• We need to handle all nine cash flow entities (government, non-energy industries, labor and consumers, foreign entities, and the five energy suppliers).
• We need to power our calculations using the holistic cash flow changes that the model already carefully calculates and sums, not changes in cash flow due to specific causes. For example, we don't want to look at how consumers spend money they keep due to energy savings, because we are going to be using data on how consumers' cash changes due to the combined effects of the entire policy package, not due to any specific component like "energy savings." Aside from being technically more tractable this way, it reflects the fact that money is fungible, and ignoring psychological quirks, people should spend an extra dollar the same way no matter the source of that extra dollar.
• We need to ensure we capture the respending of respent dollars within each year. That is, an entity that respends part of an extra dollar enriches another entity, which will go on to respend part of that dollar. We're trying to capture respending after all passes are completed in each annual timestep.
• We need to make sure we properly handle government deficit spending and that this doesn't prevent a long series of cash flow passes from converging to a finite value.
• The general approach is to complete GDP calculations first, then attempt to estimate changes in jobs/employment using "labor intensity" multipliers for different industries and entities.
• However, we need to ensure we handle economic "pie growth" effects of improved labor productivity (or capital productivity). For example, job displacements due to being able to do more work with fewer people shouldn't permanently reduce jobs, as the displaced people eventually find new jobs and overall economic productivity increases (just how productivity-increasing inventions since the Middle Ages have allowed our economy to grow without causing mass unemployment).
• We need to consider if the respending should itself increase energy use and emissions, particularly in the industrial sector, or if this is already handled properly in the first-order calculations of policy lever effects and should not be further altered.

This is a large and tricky addition that should not be undertaken lightly.

[jrissman]

Once this feature is built, look into adding related policies. For example, a policy could redirect government funding from present uses to certain other uses (e.g. direct government investment in new energy technologies, or something else). Consider other possible investment-centric policies. You could also consider implementing a government green procurement policy, per issue #54.

[jrissman]

This feature is complete and ready for internal and external review. An Input-Output (IO) model has been built inside the Energy Policy Simulator and handles macroeconomic impacts and respending. The most difficult parts were linking the direct financial outcomes calculated from the rest of the EPS to the IO model inputs, and using the IO model outputs as feedbacks so that indirect/induced spending influences demand for transportation, building services, and industrial production.

Special thanks are due to the American Council for an Energy-Efficient Economy for providing advice and guidance in constructing this IO model, including allowing me to study the workings of their IO model, DEEPER. Note that the Energy Policy Simulator uses public data from the OECD statistical database - no data from DEEPER are used in the EPS.